2s-10s is remarkably accurate for predicting recessions. For returns over 10 years or so CAPE is quite good, stocks/(stocks+bonds+cash) is very good. But you c

2s-10s is remarkably accurate for predicting recessions. For returns over 10 years or so CAPE is quite good, stocks/(stocks+bonds+cash) is very good. But you cannot make statistical claims about such numbers, the calculations are done over the entire very small population (if you want to abuse the maths a bit and call a time series a population), not over a sample.